The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code and documentation from the workspace, creating a surface for potential injection attacks where malicious content could influence the agent's behavior.\n
Ingestion points: The agent reads file contents and git diffs throughout the cleanup phases described in SKILL.md.\n
Boundary markers: There are no explicit instructions to treat code content as purely data or to ignore instructions embedded within the analyzed files.\n
Capability inventory: The skill can execute shell commands, delete files via git rm, and spawn sub-agent workers with specific briefs.\n
Sanitization: No sanitization is performed on the ingested repository content.\n- [COMMAND_EXECUTION]: The skill executes local commands to analyze and manage the repository.\n
Evidence: Instructions in SKILL.md and worker briefs utilize git, rg, and bun run check.\n- [EXTERNAL_DOWNLOADS]: The skill fetches a documentation file from a well-known service to guide its validation logic audits.\n
Evidence: SKILL.md and references/zod-audit.md instruct the agent to retrieve zod.dev/llms.txt.

unslop-code